Scientist, Model Validation and Experimental Data Analysis

TEAM AND ROLE

• Shape the architecture of the world’s first commercial fusion power plant – Own system-level decisions that determine how a first-of-a-kind energy technology is designed, integrated, and ultimately deployed at scale.

• Solve some of the most complex engineering challenges in industry – Work across tightly coupled disciplines (plasma physics, magnets, cryogenics, manufacturing, controls, and more) to resolve critical trade-offs and turn cutting-edge science into a functioning product.

• Build real hardware with a pragmatic, fast-moving team from all over the world – Combine advanced simulation and systems thinking with a strong execution mindset, focusing on practical engineering solutions that accelerate the path to commercial fusion energy.
  

WHY JOIN PROXIMA FUSION

• You will get to work on some of the most complex tech challenges to bring abundant, safe, clean energy to the world.

YOUR IMPACT

You will make Proxima's plasma simulation stack trustworthy by confronting it with reality. Every prediction we make for Alpha – Proxima's Q>1 experimental stellarator – and for Stellaris, our commercial power plant, rests on models whose credibility has to be earned against real measurements from an operating device.

W7-X is where we earn it. You will take our integrated modeling framework, FUS3, and our Bayesian plasma state estimation workflow to W7-X data: reconstruct the plasma state from diagnostic measurements, compare our predictive models against what the machine actually did, and turn the discrepancies into a prioritized map of where our physics is right, where it is not, and what we must fix before Alpha's first plasma. Modeling sits on Alpha's operational critical path and cannot be backfilled after first plasma – this role is how we get ahead of it.

You will be the person in the group who lives at the intersection of modeling and experiment: fluent enough in the physics to know what a model should predict, and close enough to the diagnostics to know what the data can actually support.

WHAT YOU WILL DO

• Validate Proxima's predictive plasma models against W7-X experimental data – neoclassical and turbulent transport, 3D equilibrium, ECRH deposition, pellet fueling and drift, bootstrap current and iota evolution – and record each comparison as a reproducible validation report with quantitative agreement metrics, stated limitations, and a pinned regression test

• Reconstruct the plasma state from real diagnostic measurements using our time-dependent Bayesian state-estimation workflow (Thomson scattering, interferometry, CXRS, ECE, XICS, bolometry, magnetics), with calibrated uncertainties, and benchmark it against established analyses such as Minerva/IDA reconstructions and ASTRA power-balance results

• Work with raw and pre-processed diagnostic data, not just analyzed profiles – instrument geometry, sightlines, calibrations and the caveats that come with them – and own the forward models, noise models and profile fits that feed transport analysis

• Validate against databases of discharges rather than single shots, across W7-X magnetic configurations and operating regimes, so the output is a map of model applicability and parameter trends rather than one-off discrepancy reports

• Act as a technical interface to W7-X and IPP: scope and prioritize data requests, spend time on site in Greifswald working alongside diagnosticians and operators, and propose and support experiments that directly target our model gaps

• Feed validation back into design – convert diagnostic placement and observability studies on W7-X into requirements for Alpha's diagnostic set, and into realistic uncertainty models for the synthetic measurements we use to design against

• Extend the modeling stack itself: mappers from experimental archives into IMAS, synthetic diagnostics, and analysis workflows, so that interpreting a shot end-to-end – raw data to reconstructed plasma state to physics quantities – runs in minutes and is reproducible by anyone in the group

• As Alpha approaches operation, help stand up the between-shot analysis loop that will run on day one

WHO YOU ARE

• PhD in plasma physics, fusion science, or a closely related discipline

• Hands-on experience with experimental data from a magnetic confinement device – you have dealt with diagnostics, calibrations, systematic errors, and all the ways real data is messier than simulation output

• Experience reconstructing the plasma state from measurements: profile fitting, integrated data analysis, equilibrium reconstruction, or power and particle balance analysis

• Strong grounding in transport, equilibrium, and heating physics – enough to tell a code bug from a physics discovery

• Rigorous about uncertainty; you treat a validation claim without error bars on both sides as incomplete

• Comfortable writing and maintaining scientific software – version control, testing, and reproducibility are habits, not overhead

• Can communicate effectively with physicists, diagnosticians, engineers, and experimentalists, and work productively inside an external collaboration

• Self-directed; able to identify what needs to be done and execute

• Willing to travel regularly to Greifswald, and later to the Alpha site

NICE TO HAVE

• W7-X experience - operations, diagnostics, or data analysis, and familiarity with the W7-X data archive and Minerva

• Stellarator physics background

• Bayesian inference, uncertainty quantification, or machine learning applied to experimental data

• Experience with integrated modeling and transport codes (FUSE, OMFIT, ASTRA, TRANSP, FUSE, JINTRAC, or similar) and with neoclassical (SFINCS, DKES, NEO) or gyrokinetic (GENE, CGYRO, GX, or similar) codes

• Julia and Python programming – our integrated modeling framework is written in Julia

• Familiarity with the IMAS data model

• Familiarity with AI-assisted development tools

• A track record of published validation studies or successful experimental proposals

INTERVIEW PROCESS

• Recruiter Interview (30-60 min)

• Technical Screening (30 min)

• Technical Panel (3x60 min)